A digital camera has human extraction means, non-human extraction means, and composition judgment means. The human extraction means extracts a human figure region by analysis of image data. The non-human extraction means extracts a major subject other than a human figure by analysis of a region other than the human figure region having been extracted by the human extraction means. The composition judgment means evaluates arrangement of the human figure and the major subject according to results of the extraction, and judges whether composition is appropriate. Based on whether the composition is appropriate, timing to record the image data is determined. Preferably, recording means is controlled so as to record the image data at the determined timing, or the timing is notified to a user.